GS-GPU-04Engineering datasheet

NVIDIA B300

Statusshipping
Verified2026-07-10

The B300 is NVIDIA's Blackwell Ultra data center GPU, a binned and re-engineered Blackwell die that trades the B200's 8-Hi HBM3e stacks for 12-Hi stacks to land 288 GB per GPU, and raises the power ceiling to 1,400 W to push dense NVFP4 throughput about 1.5x past B200. It ships on the same 8-GPU HGX baseboard pattern as B200 and is also the GPU building block of the GB300 NVL72 rack (72 GPUs, 36 Grace CPUs, fully liquid-cooled). In a private AI build it is the part to reach for when a single GPU's memory ceiling, not raw FLOPS, is what is blocking a large-context or reasoning-model deployment. As of mid-2026 it is NVIDIA's current top-of-stack Blackwell part, shipping since January 2026, and new OEM quotes are increasingly steering toward it over B200.

288 GB HBM3e
Memory · 8 TB/s bandwidth, 12-Hi stacks
15 / 20 PFLOPS
FP4 Tensor (NVFP4) · dense / sparse, per GPU
5 / 10 PFLOPS
FP8 Tensor · dense / sparse, per GPU
up to 1400 W
TGP · requires direct liquid cooling
1.8 TB/s bidirectional
NVLink · NVLink 5, per GPU
Compute01/05
ArchitectureBlackwell Ultra, dual-die · 208 billion transistors across both dies
Streaming Multiprocessors160 SMs · 8 graphics processing clusters
Tensor Cores640 · 5th generation
FP8 Tensor5 / 10 PFLOPS · dense / sparse
FP4 Tensor (NVFP4)15 / 20 PFLOPS · dense / sparse, 1.5x B200 dense
Attention throughput~2x B200 · doubled SFU throughput for attention layers
Memory02/05
Capacity288 GB HBM3e · 50% more than B200's 180 GB
Stack configuration8 stacks, 12-Hi · vs B200's 8-Hi stacks
Bandwidth8 TB/s · per GPU
Bus width8,192-bit · 16 x 512-bit memory controllers
Interconnect03/05
NVLink1.8 TB/s bidirectional · NVLink 5, same generation as B200
Die-to-die (NV-HBI)10 TB/s · internal, between the two dies
PCIePCIe Gen6, 256 GB/s · double B200's Gen5 host link
NVLink-C2C (Grace pairing)900 GB/s · coherent CPU-GPU link in GB300 Superchip
GB300 NVL72 rack fabric130 TB/s aggregate · 72-GPU NVLink domain
Power & thermal04/05
TGPup to 1400 W · per GPU, ~40% above B200
8-GPU baseboard power~11.2 kW · GPUs only, before CPUs and networking
Coolingdirect liquid cooling required · no air-cooled HGX B300 or DGX B300 SKU
DGX B300 system power~14 kW · 8-GPU system including CPUs, DPUs, storage
Physical & platform05/05
Form factorSXM module · not sold as a discrete PCIe card
BaseboardHGX B300 8-GPU · 2.3 TB aggregate HBM3e
GB300 NVL7272 GPUs + 36 Grace CPUs · 1 rack, fully liquid-cooled
GB300 NVL72 FP4 (rack)~1.1 exaFLOPS dense · aggregate, sparse figure runs higher
Reference systemsDGX B300, GB300 NVL72, OEM HGX B300 · Supermicro, Dell, HPE, Lenovo qualifying
Field notes
  • The FP4 sparse-to-dense ratio on B300 is only about 1.33x (15 to 20 PFLOPS), not the 2x you'd expect from structured sparsity and that B200 actually delivers (9 to 18 PFLOPS). Do not assume Blackwell Ultra scales sparsity the same way standard Blackwell does when you're modeling throughput.
  • 1,400 W per GPU with no air-cooled option means the facility conversation (liquid loop, CDU capacity, floor loading) has to happen before you place the GPU order, not after. This is a harder requirement than B200, where air-cooled HGX chassis exist.
  • The jump to 288 GB matters most when a single model replica, its KV cache, or its optimizer state doesn't fit in 180 GB. If your working set fits comfortably on B200, the memory headroom on B300 buys you little beyond the ~1.5x dense FP4 uplift.
  • H200 (141 GB HBM3e, air-cooled, Hopper software stack) is still the pragmatic choice for teams that need memory-bandwidth-bound inference today and cannot absorb a liquid-cooling retrofit for B300's 1,400 W ceiling.
  • Treat published NVFP4 throughput as a ceiling, not a delivered number: it assumes a quantization pipeline that actually produces valid NVFP4 weights and a kernel stack tuned for it. Benchmark your own model before sizing a cluster on the PFLOPS line.
Questions we get on this part

How much does an NVIDIA B300 cost?

Reported figures put a single B300 GPU around $50,000, with a fully configured 8-GPU DGX B300 system in the $300,000 to $500,000 range. A full GB300 NVL72 rack has been estimated near $3.7 to $4.0 million based on large hyperscaler order commentary. These are reported market figures, not published NVIDIA list prices.

B200 vs B300: what's the actual difference?

B300 has 288 GB of HBM3e versus B200's 180 GB, a 1,400 W power ceiling versus B200's 1,000 W, and roughly 1.5x B200's dense NVFP4 throughput per GPU. B300 also requires direct liquid cooling in every deployment path, while B200 has air-cooled HGX options.

Does B300 require liquid cooling?

Yes. At up to 1,400 W per GPU there is no air-cooled HGX B300 or DGX B300 SKU; every reference platform (HGX B300, DGX B300, GB300 NVL72) ships as direct-liquid-cooled.

What is GB300 NVL72?

GB300 NVL72 is NVIDIA's rack-scale system built from B300: 72 Blackwell Ultra GPUs and 36 Grace CPUs in one liquid-cooled rack, connected by a 130 TB/s aggregate NVLink fabric, delivering roughly 1.1 exaFLOPS of dense FP4 throughput at the rack level.